giscanner/scannerlexer.l

   1 /* -*- Mode: C -*-
   2  * GObject introspection: C lexer
   3  *
   4  * Copyright (c) 1997 Sandro Sigala  <ssigala@globalnet.it>
   5  * Copyright (c) 2007-2008 Jürg Billeter  <j@bitron.ch>
   6  *
   7  * All rights reserved.
   8  *
   9  * Redistribution and use in source and binary forms, with or without
  10  * modification, are permitted provided that the following conditions
  11  * are met:
  12  * 1. Redistributions of source code must retain the above copyright
  13  *    notice, this list of conditions and the following disclaimer.
  14  * 2. Redistributions in binary form must reproduce the above copyright
  15  *    notice, this list of conditions and the following disclaimer in the
  16  *    documentation and/or other materials provided with the distribution.
  17  *
  18  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
  19  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
  20  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
  21  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
  22  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
  23  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  24  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  25  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  26  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
  27  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  28  */
  29
  30 %{
  31 #include <ctype.h>
  32 #include <stdio.h>
  33
  34 #include <glib.h>
  35 #include "sourcescanner.h"
  36 #include "scannerparser.h"
  37 #include "grealpath.h"
  38
  39 int lineno;
  40
  41 #undef YY_BUF_SIZE
  42 #define YY_BUF_SIZE 1048576
  43
  44 extern int yylex (GISourceScanner *scanner);
  45 #define YY_DECL int yylex (GISourceScanner *scanner)
  46 static int yywrap (void);
  47 static void parse_comment (GISourceScanner *scanner);
  48 static void process_directive (GISourceScanner *scanner);
  49 static int check_identifier (GISourceScanner *scanner, const char *);
  50 static int parse_ignored_macro (void);
  51 %}
  52
  53 %option nounput
  54
  55 intsuffix                               ([uU][lL]?[lL]?)|([lL][lL]?[uU]?)
  56 fracconst                               ([0-9]*\.[0-9]+)|([0-9]+\.)
  57 exppart                                 [eE][-+]?[0-9]+
  58 floatsuffix                             [fFlL]
  59 chartext                                ([^\\\'])|(\\.)
  60 stringtext                              ([^\\\"])|(\\.)
  61
  62 %%
  63
  64 "\n"                                    { ++lineno; } /* " */
  65 "\\\n"                                  { ++lineno; }
  66 [\t\f\v\r ]+                            { /* Ignore whitespace. */ }
  67
  68 "/*"                                    { parse_comment(scanner); }
  69 "//".*                                  { }
  70
  71 "#define "[a-zA-Z_][a-zA-Z_0-9]*"("     { yyless (yyleng - 1); return FUNCTION_MACRO; }
  72 "#define "[a-zA-Z_][a-zA-Z_0-9]*        { return OBJECT_MACRO; }
  73
  74 "#"                                     { process_directive(scanner); }
  75
  76 "{"                                     { return '{'; }
  77 "<%"                                    { return '{'; }
  78 "}"                                     { return '}'; }
  79 "%>"                                    { return '}'; }
  80 "["                                     { return '['; }
  81 "<:"                                    { return '['; }
  82 "]"                                     { return ']'; }
  83 ":>"                                    { return ']'; }
  84 "("                                     { return '('; }
  85 ")"                                     { return ')'; }
  86 ";"                                     { return ';'; }
  87 ":"                                     { return ':'; }
  88 "..."                                   { return ELLIPSIS; }
  89 "?"                                     { return '?'; }
  90 "."                                     { return '.'; }
  91 "+"                                     { return '+'; }
  92 "-"                                     { return '-'; }
  93 "*"                                     { return '*'; }
  94 "/"                                     { return '/'; }
  95 "%"                                     { return '%'; }
  96 "^"                                     { return '^'; }
  97 "&"                                     { return '&'; }
  98 "|"                                     { return '|'; }
  99 "~"                                     { return '~'; }
 100 "!"                                     { return '!'; }
 101 "="                                     { return '='; }
 102 "<"                                     { return '<'; }
 103 ">"                                     { return '>'; }
 104 "+="                                    { return ADDEQ; }
 105 "-="                                    { return SUBEQ; }
 106 "*="                                    { return MULEQ; }
 107 "/="                                    { return DIVEQ; }
 108 "%="                                    { return MODEQ; }
 109 "^="                                    { return XOREQ; }
 110 "&="                                    { return ANDEQ; }
 111 "|="                                    { return OREQ; }
 112 "<<"                                    { return SL; }
 113 ">>"                                    { return SR; }
 114 "<<="                                   { return SLEQ; }
 115 ">>="                                   { return SREQ; }
 116 "=="                                    { return EQ; }
 117 "!="                                    { return NOTEQ; }
 118 "<="                                    { return LTEQ; }
 119 ">="                                    { return GTEQ; }
 120 "&&"                                    { return ANDAND; }
 121 "||"                                    { return OROR; }
 122 "++"                                    { return PLUSPLUS; }
 123 "--"                                    { return MINUSMINUS; }
 124 ","                                     { return ','; }
 125 "->"                                    { return ARROW; }
 126
 127 "__asm"                                 { if (!parse_ignored_macro()) REJECT; }
 128 "__asm__"                               { if (!parse_ignored_macro()) REJECT; }
 129 "__attribute__"                         { if (!parse_ignored_macro()) REJECT; }
 130 "__attribute"                           { if (!parse_ignored_macro()) REJECT; }
 131 "__const"                               { return CONST; }
 132 "__extension__"                         { return EXTENSION; }
 133 "__inline"                              { return INLINE; }
 134 "__nonnull"                             { if (!parse_ignored_macro()) REJECT; }
 135 "__signed__"                            { return SIGNED; }
 136 "__restrict"                            { return RESTRICT; }
 137 "__typeof"                              { if (!parse_ignored_macro()) REJECT; }
 138 "_Bool"                                 { return BOOL; }
 139
 140 [a-zA-Z_][a-zA-Z_0-9]*                  { if (scanner->macro_scan) return IDENTIFIER; else REJECT; }
 141
 142 "asm"                                   { if (!parse_ignored_macro()) REJECT; }
 143 "auto"                                  { return AUTO; }
 144 "break"                                 { return BREAK; }
 145 "case"                                  { return CASE; }
 146 "char"                                  { return CHAR; }
 147 "const"                                 { return CONST; }
 148 "continue"                              { return CONTINUE; }
 149 "default"                               { return DEFAULT; }
 150 "do"                                    { return DO; }
 151 "double"                                { return DOUBLE; }
 152 "else"                                  { return ELSE; }
 153 "enum"                                  { return ENUM; }
 154 "extern"                                { return EXTERN; }
 155 "float"                                 { return FLOAT; }
 156 "for"                                   { return FOR; }
 157 "goto"                                  { return GOTO; }
 158 "if"                                    { return IF; }
 159 "inline"                                { return INLINE; }
 160 "int"                                   { return INT; }
 161 "long"                                  { return LONG; }
 162 "register"                              { return REGISTER; }
 163 "restrict"                              { return RESTRICT; }
 164 "return"                                { return RETURN; }
 165 "short"                                 { return SHORT; }
 166 "signed"                                { return SIGNED; }
 167 "sizeof"                                { return SIZEOF; }
 168 "static"                                { return STATIC; }
 169 "struct"                                { return STRUCT; }
 170 "switch"                                { return SWITCH; }
 171 "typedef"                               { return TYPEDEF; }
 172 "union"                                 { return UNION; }
 173 "unsigned"                              { return UNSIGNED; }
 174 "void"                                  { return VOID; }
 175 "volatile"                              { return VOLATILE; }
 176 "while"                                 { return WHILE; }
 177
 178 [a-zA-Z_][a-zA-Z_0-9]*                  { return check_identifier(scanner, yytext); }
 179
 180 "0"[xX][0-9a-fA-F]+{intsuffix}?         { return INTEGER; }
 181 "0"[0-7]+{intsuffix}?                   { return INTEGER; }
 182 [0-9]+{intsuffix}?                      { return INTEGER; }
 183
 184 {fracconst}{exppart}?{floatsuffix}?     { return FLOATING; }
 185 [0-9]+{exppart}{floatsuffix}?           { return FLOATING; }
 186
 187 "'"{chartext}*"'"                       { return CHARACTER; }
 188 "L'"{chartext}*"'"                      { return CHARACTER; }
 189
 190 "\""{stringtext}*"\""                   { return STRING; }
 191 "L\""{stringtext}*"\""                  { return STRING; }
 192
 193 .                                       { if (yytext[0]) fprintf(stderr, "%s:%d: unexpected character `%c'\n", scanner->current_filename, lineno, yytext[0]); }
 194
 195 %%
 196
 197 static int
 198 yywrap (void)
 199 {
 200   return 1;
 201 }
 202
 203
 204 static void
 205 parse_comment (GISourceScanner *scanner)
 206 {
 207   GString *comment;
 208   int c1, c2;
 209
 210   c1 = input();
 211   c2 = input();
 212
 213   comment = g_string_new ("");
 214
 215   while (c2 != EOF && !(c1 == '*' && c2 == '/'))
 216     {
 217       g_string_append_c (comment, c1);
 218
 219       c1 = c2;
 220       c2 = input();
 221
 222     }
 223
 224   scanner->comments = g_slist_prepend (scanner->comments,
 225                                        g_string_free (comment, FALSE));
 226 }
 227
 228 static int
 229 check_identifier (GISourceScanner *scanner,
 230                   const char  *s)
 231 {
 232         /*
 233          * This function checks if `s' is a type name or an
 234          * identifier.
 235          */
 236
 237         if (gi_source_scanner_is_typedef (scanner, s)) {
 238                 return TYPEDEF_NAME;
 239         } else if (strcmp (s, "__builtin_va_list") == 0) {
 240                 return TYPEDEF_NAME;
 241         }
 242
 243         return IDENTIFIER;
 244 }
 245
 246 static void
 247 process_directive (GISourceScanner *scanner)
 248 {
 249         /* extract current filename from #line directives */
 250         GString *filename_builder;
 251         gboolean in_string, found_filename;
 252
 253         lineno = 0;
 254         found_filename = FALSE;
 255         in_string = FALSE;
 256         filename_builder = g_string_new ("");
 257
 258         int c = input ();
 259         while (c != EOF && c != '\n') {
 260                 if (!in_string) {
 261                         if (c == '\"') {
 262                                 in_string = TRUE;
 263                                 found_filename = TRUE;
 264                         } else if (c >= '0' && c <= '9') {
 265                                 if (!found_filename) {
 266                                         lineno = lineno * 10 + (c - '0');
 267                                 }
 268                         }
 269                 } else {
 270                         if (c == '\"') {
 271                                 in_string = FALSE;
 272                         } else if (c == '\\') {
 273                                 g_string_append_c (filename_builder, c);
 274                                 c = input ();
 275                                 g_string_append_c (filename_builder, c);
 276                         } else {
 277                                 g_string_append_c (filename_builder, c);
 278                         }
 279                 }
 280                 c = input ();
 281         }
 282
 283         if (filename_builder->len > 0) {
 284                 char *filename = g_strcompress (filename_builder->str);
 285                 if (g_realpath (filename))
 286                   {
 287                     g_free (scanner->current_filename);
 288                     scanner->current_filename = g_realpath (filename);
 289                     g_assert (scanner->current_filename);
 290                     g_free(filename);
 291                   }
 292         }
 293
 294         g_string_free (filename_builder, TRUE);
 295 }
 296
 297 /*
 298  * This parses a macro which is ignored, such as
 299  * __attribute__((x)) or __asm__ (x)
 300  */
 301 static int
 302 parse_ignored_macro (void)
 303 {
 304         int c;
 305         int nest;
 306
 307         while ((c = input ()) != EOF && isspace (c))
 308                 ;
 309         if (c != '(')
 310                 return FALSE;
 311
 312         nest = 0;
 313         while ((c = input ()) != EOF && (nest > 0 || c != ')')) {
 314                 if (c == '(')
 315                         nest++;
 316                 else if (c == ')')
 317                         nest--;
 318                 else if (c == '"') {
 319                         while ((c = input ()) != EOF && c != '"') {
 320                                 if (c == '\\')
 321                                         c = input ();
 322                         }
 323                 } else if (c == '\'') {
 324                         c = input ();
 325                         if (c == '\\')
 326                                 c = input ();
 327                         else if (c == '\'')
 328                                 return FALSE;
 329                         c = input ();
 330                         if (c != '\'')
 331                                 return FALSE;
 332                 } else if (c == '\n')
 333                         lineno++;
 334         }
 335
 336         return TRUE;
 337 }